search

bgruening / docker-jupyter-notebook

:whale: :microscope: :books: Jupyter running in a docker container. This image can be used to integrate Jupyter into Galaxy
MIT License
35 stars 39 forks source link

Update jupyter/datascience-notebook #52

Closed Marie59 closed 1 year ago

Marie59 commented 1 year ago

Change the tag of jupyter/datascience-notebook to the one with julia-1.8.5. There are even more recent releases but I did not no if it was relevant to take one of them. This is all quite new for me but if I understood correctly I think it's the only change that need to be done in order to update the julia version. @bgruening tell me if I need to do anything else.

Marie59 commented 1 year ago

Isn't it possible to replace RUN jupyter labextension install @jupyterlab/geojson-extension by RUN conda install jupyterlab-geojson ? and do the same for the following packages ? Could it fix the issue ? (maybe I am saying nonsense here sorry if it's the case)

bgruening commented 1 year ago

I don't know @Marie59, we need to test. Does it work if you build it locally? I'm just trying here :)

Marie59 commented 1 year ago

I don't know @Marie59, we need to test. Does it work if you build it locally? I'm just trying here :)

I am doing something wrong to build the image locally because I can't build any image locally (but I can run them). I am trying to fix that on my machine.

Marie59 commented 1 year ago

Okay @bgruening the build on my machine of jupyter/datascience-notebook finally worked ! But the build of this docker-jupyter-notebook did not and broke with the same error as here so I'm going to test what i suggested yesterday (it seems to work on my machine).

bgruening commented 1 year ago

@Marie59 its green :)

But we still don't know if this extension works in Jupyter so this would need to be tested locally. And we need to include the other extensions as well.

Marie59 commented 1 year ago

@Marie59 its green :)

But we still don't know if this extension works in Jupyter so this would need to be tested locally. And we need to include the other extensions as well.

Yes ! I will try to see how to test locally. Do you mean that for the other extensions I should also change to RUN conda install XXX ?

bgruening commented 1 year ago

Yes, if this works as we hope, we should use conda for all of them.

Marie59 commented 1 year ago

ok then I think one of them is missing on conda the jupyterlab/toc-extension.

Marie59 commented 1 year ago

Ah no just found it ! forget what I said

bgruening commented 1 year ago

https://github.com/jupyterlab/jupyterlab-toc

but it also looks dead? Skip it if it makes problems and feel free to add others if useful :)

Marie59 commented 1 year ago

So, I removed jupyterlab-toc because it is now included in jupyterlab. I try to build and then run the image using RUN conda install for the 3 extensions but with keeping the old FROM jupyter/datascience-notebook:d990a62010ae everything is working. I don't know if I need to test something else to check if the extensions are working ? However, the new FROM jupyter/datascience-notebook:python-3.10 is building correctly but not running correctly due to a migration to Notebook 7. I am looking into it but if you have any leads on that...

bgruening commented 1 year ago

Mh, no I don't really know :-( Maybe that is helpful? https://jupyterlab-contrib.github.io/migrate_from_classical.html

Marie59 commented 1 year ago

Okay I was able to run it. I am pretty sure the extension katex and geojson are working fine wwith the migration. I am not sure for the fasta extension. I have more difficulties to find a way to test it (I will keep searching to be sure during the day). However I found that https://github.com/jupyterlab/jupyter-renderers I think it could come replace fasta and geojson extensions but it's not on conda yet. Do you think I should try to make a conda recipe for that (never try for a jupyter package) ?

bgruening commented 1 year ago

If you think this is a good package, we can create a conda one for it.

grayscull pypi jupyter-renderers should generate us a conda-forge package.

Marie59 commented 1 year ago

It does not seem to be on pypi. But, after checking it is called in the recipes of the fasta and geojson extension here home: https://github.com/jupyterlab/jupyter-renderers. So maybe, the packages on conda are indeed up to date with that (I am not quite sure here)

Marie59 commented 1 year ago

Yep I confirm everything is linked back to jupyter-renderers forget what I said sorry !

bgruening commented 1 year ago

Sorry, that this was closed automatically. I think Github was confused because you used your master branch.

I created a new container: https://quay.io/repository/bgruening/docker-jupyter-notebook?tab=tags

23.3 should be the new one, with your changes :)

I think you can now use this new container and just add your notebooks and "done" :)

Thanks!